Google v. Oracle: Takeaways for Software Preservation, Cultural Heritage, and Fair Use Generally

Prepared for the Software Preservation Network by Brandon Butler

On April 5, 2021, the Supreme Court issued its opinion on the long-running litigation between Oracle and Google over the reuse of aspects of Oracle’s Java programming framework in Google’s Android mobile operating system. The majority opinion, written by Justice Breyer and joined by five of his fellow justices (Chief Justice Roberts, and Justices Kagan, Sotomayor, Kavanaugh, and Gorsuch), sided with Google, saying its use was lawful because it was protected by fair use. Justice Thomas wrote a dissent, joined only by Justice Alito, arguing that Google’s use was infringing. The newest Justice, Amy Coney Barrett, did not participate in the arguments or decision of the case as it predated her joining the Court. More background on the case can be found in my earlier blog post for SPN summarizing the oral arguments.

Justice Breyer’s opinion is already a landmark for the reasons I laid out there: it is the first Supreme Court opinion to address fair use in nearly thirty years—the last one was Campbell v. Acuff-Rose in 1994. And it is the first Supreme Court opinion to address copyright’s protection for software—ever. And now we know that the opinion will be a milestone for another reason: it is a confident, erudite treatment of the issue by a Justice who has been thinking about copyright and software for more than half a century. As a law professor, Stephen Breyer earned tenure at Harvard based on his 1970 article, “The Uneasy Case for Copyright: A Study of Copyright in Books, Photocopies, and Computer Programs.” The opinion is thus a very happy coincidence: a thorny and consequential issue confronted by a subtle and experienced thinker. The results are quite encouraging for software preservation and for cultural heritage institutions and fair users generally.

Software Preservation Takeaways

The Oracle opinion includes a laundry list of insights into copyright’s application to software, and in particular to fair use’s role in software re-use. While we should be careful not to over-read the victory in this case, there is much to celebrate here.

This case is not just about reimplementing APIs

First, we should note and reject the effort from copyright industry lobbyists to minimize the Oracle opinion and characterize it as limited to the narrow context of this case (the reimplementation of API interface elements, to state it as narrowly as possible). Every judicial opinion is two things at once: it decides the case in front of it, and it reviews, announces, or clarifies the legal principles involved in reaching that decision. This is especially true for Supreme Court opinions, which are almost always the result of the Court perceiving a legal issue that needs clarifying so that other courts can reach better decisions in similar cases. While Justice Breyer says that his opinion is not intended to “overturn or modify” the court’s previous fair use rulings, he also says that the court has reached its opinion by applying general fair use principles. In doing so, the Court necessarily elucidates these principles and gives new insight into how they could apply to future cases, especially future software cases. So, while this is a case about reimplementing an API, it is not just a case about that.

Fair use applies equally to software (as do all the other limitations and exceptions in copyright).

The majority opinion assumes that the code at issue is protected (a question that was raised in the case, but that the court chose not to settle), but promptly reminds us that while “owners of computer programs enjoy the exclusive rights” in the Copyright Act, those rights “are limited like any other works” by the limitations and exceptions in the law. In particular, “[j]ust as fair use distinguishes among books and films, which are indisputably subjects of copyright, so too must it draw lines among computer programs.” Recognizing copyright’s limitations as well as its exclusive rights is a key part of remaining “faithful to the Copyright Act’s overall design.” It may seem banal, but this observation is important because there are plenty of folks who would argue (as Justice Thomas nearly does in his dissent) that copyright protection for software is not subject to the limits that apply to other kinds of works.

Fair use is especially important for software

Justice Breyer quotes approvingly the observation that “applying copyright law to computer programs is like assembling a jigsaw puzzle whose pieces do not quite fit,” because computer programs have a strong functional aspect unlike anything found in books, film scripts, and other “literary works” protected by copyright. The law treats functional things with more suspicion in terms of granting monopolies—they typically have to go through the more expensive and onerous patent process, and the monopoly that results is much shorter, so that useful discoveries can pass more quickly into unfettered public use. The congressional commission that examined whether to include software in copyright’s scope (rather than leave it solely to patent) was “Mindful of not ‘unduly burdening users of programs and the general public,’ [and] wrote that copyright ‘should not grant anyone more economic power than is necessary to achieve the incentive to create.'”

With this policy objective in mind, Justice Breyer argues that fair use has a crucial role to play in keeping software copyright in balance:

The upshot, in our view, is that fair use can play an important role in determining the lawful scope of a computer program copyright, such as the copyright at issue here…. It can focus on the legitimate need to provide incentives to produce copyrighted material while examining the extent to which yet further protection creates unrelated or illegitimate harms in other markets or to the development of other products. In a word, it can carry out its basic purpose of providing a context-based check that can help to keep a copyright monopoly within its lawful bounds. [emphasis added]

Well! All this certainly suggests that the software preservation community was wise to invest time and resources in developing and promulgating a consensus view of fair use applied to software preservation, as fair use must play a central role in regulating the scope of software copyright.

It is much easier now to explain why software reuse for preservation, teaching, and research is transformative

In deciding whether a use is fair, courts (very) strongly favor uses that are found to be “transformative,” a legal term of art that describes uses “at the heart” of fair use’s protection because they advance copyright’s Constitutional purpose. In Justice Breyer’s words, a transformative use does something “important and new” relative to the original work used. Sometimes this involves literally changing the work into something different (as in a parody song that uses the original melody but new lyrics that mock the original), but just as often a work is used unchanged but for a new purpose (as in the creation of a search database for images on the internet).

Deciding whether a use is “new” can depend on how the nature and the purpose of the use is characterized. In this case, Google’s use can be described in a way that makes it sound hardly new at all. As Justice Breyer writes, “Google copied portions of the Sun Java API precisely, and it did so in part for the same reason that Sun created those portions, namely, to enable programmers to call up implementing programs that would accomplish particular tasks.”

Nevertheless, Justice Breyer warns that “since virtually any unauthorized use of a copyrighted computer program (say, for teaching or research) would do the same, to stop here would severely limit the scope of fair use in the functional context of computer programs.” Two remarkable and important things happen in this one sentence. First, it’s clear that severely limiting the scope of fair use is not an acceptable outcome. Others (like Justice Thomas in dissent) might be perfectly comfortable with that outcome, but the majority makes clear that such a miserly approach is not tenable under the law. If your approach to transformative use puts it out of reach in the vast majority of cases, then your approach is wrong.

Second, Justice Breyer calls out “teaching and research” as core fair use purposes that had better not be severely limited by an overly simplistic analysis of transformative use. That is, the fact that a narrow, literal reading of transformative use would chill teaching and research is itself a reason to reject that reading. To avoid this outcome, “we must go further and examine the copying’s more specifically described ‘purpose[s]’ and ‘character.'”

To go further, Justice Breyer explains that Google’s Android platform for smartphone app developers was a new and important product that incorporated Java code only insofar as it was needed to ease the way for developers moving to the new platform. He quotes from amicus briefs and refers to the trial record evidence showing that reusing code in this particular context (reusing declaring code) without permission is a normal way of creating new products, is necessary for interoperability, is necessary to enable programmers to use their skills (including familiarity with Java API calls), and is widely believed to be a salutary aspect of software creation, which is in turn supportive of the Constitutional purpose of copyright – to “promote Progress.”

The upshot of all this for software preservation and scholarly access is that it creates a clear pathway for practitioners to explain why their uses are transformative, and to push back against overly-simplistic characterizations that would eviscerate fair use. Consider, for example, an institution that makes software in its collection available to scholars in an emulated environment so that researchers can access archival material stored in proprietary file formats. One characterization of that use is that the software is being copied and made accessible for the very same reason it was originally created—to open and read/manipulate files. But Justice Breyer has warned us against such a simple, generic characterization, and urged us to go further and look more closely in order to ensure we preserve breathing space for fair use—especially for teaching and research.

Accordingly, we can more specifically describe the nature and purpose(s) of software reuse in a research setting to show how it is fundamentally different from ordinary consumer use: it ensures continuing access to historic material for scholarly reference and teaching, furthering the development of scholarship about archival materials (including software itself), enabling scholars and teachers to use their acquired skills (familiarity with archival materials), and ultimately promoting progress. Just like Google’s reimplementation of Java code, cultural heritage institutions’ reuse of old software in research-oriented emulation environments enables a “new collection of tasks” (namely, research and teaching tasks) operating in a “distinct and different computing environment” (namely, in an emulated environment on modern, networked hardware).

It is fair to copy entire software works if it’s appropriate to your valid, transformative purpose

This is not a new principle, but it is reaffirmed in Justice Breyer’s discussion of the third factor – amount and substantiality of the work used. It is a common misconception that the third fair use factor—amount—is a simple sliding scale, and that the more of a work you use, the less the law will favor you. In fact, fair use protects a wide variety of uses, including uses of entire works. How can this be? As Justice Breyer explains, quoting from Campbell v. Acuff-Rose,”the extent of permissible copying varies with the purpose and character of the use.” Since transformative purposes are strongly favored by fair use, “The “substantiality” factor will generally weigh in favor of fair use where, as here, the amount of copying was tethered to a valid, and transformative, purpose.”

For software preservation, the appropriate amount for preservation purposes is almost always the entire work. Similarly, research and teaching uses will typically require access to entire works (though that access can be limited and tailored as appropriate, and time-limited, so providing emulated access is arguably modest compared to providing a copy). The old “less is better, more is worse” myth can make people nervous about this, but Justice Breyer’s opinion here should help give folks confidence to use the amount appropriate to their purpose, up to and including entire works.

The approach to market analysis in Oracle should strongly favor preservation of out-of-commerce software.

The final factor in the fair use statute is the effect of the new use on the market for the work, and Justice Breyer’s approach to this factor opens interesting doors for software preservation. Justice Breyer considers three sub-factors:

  1. The amount of revenue a rights holder stands to lose from a use (including widespread similar uses by others).
  2. The source of the loss (i.e., why will the use cause revenue loss, if any).
  3. The public benefits of the use and their importance relative to the amount of revenue loss.

In considering these, the court finds that each one favors Google, and parallel arguments could be made regarding software preservation.

There’s no actual loss of revenue

Looking at the evidence from the trial, Justice Breyer concludes that the jury (which is responsible for finding the facts in a specific case) could reasonably have concluded that Google’s use did not cause any revenue loss for Oracle. This is because Sun/Oracle was “poorly positioned to succeed” in the new smartphone market that Google developed with Android. Accordingly, Google’s use of Java did not unfairly compete with Java in that market, or displace it from the market.

Similarly, software vendors generally are not positioned to support research use of out-of-commerce software (especially software that cannot be run on contemporary hardware). They are focused on developing, marketing, and supporting products for the consumer market and contemporary hardware. Indeed part of their strategy in this process is to replace old products with new ones, encouraging users to upgrade or switch by enticing them with new features, compatibility with the latest hardware, etc., and eventually ending support for older versions. Preservation and research access are no threat to the vendors’ consumer market, but serve only the “new” market for research access, which the vendors (like Oracle) cannot support. Hence, preservation and research uses cause no lost revenue for software vendors.

The nature of the loss, if any, would be due to lock-in, not the rightsholder’s creativity

Although Sun/Oracle couldn’t have competed with Google in the smartphone market, perhaps they could have profited by demanding a license from Google for use of Java. What is the nature of this lost revenue – i.e., is it akin to revenue lost due to substitution (which weighs against fair use), or is it more like revenue lost due to a negative review (which does not)? The core policy insight that fuels Justice Breyer’s opinion is that the main source of value for Java’s API elements is the time and effort that developers have invested in learning to use them. In oral arguments, Justice Breyer compared the API to the QWERTY keyboard layout—it’s not an inherently valuable or useful system, but once enough users have learned it and come to rely on it, it’s extremely valuable and important that future keyboards/development frameworks include it.

So, Justice Breyer concludes, “This source of Android’s profitability has much to do with third parties’ (say, programmers’) investment in Sun Java programs. It has correspondingly less to do with Sun’s investment in creating the Sun Java API. We have no reason to believe that the Copyright Act seeks to protect third parties’ investment in learning how to operate a created work.”

Institutions engaged in software preservation find themselves in much the same position as Google: they are tasked with building something (here a research collection of preserved software) that addresses requirements (opening legacy files, researching legacy software titles) that result primarily from the historic investments of third parties in learning and using certain software titles, not on the creative virtues of the titles themselves. Keyboard makers need QWERTY because of typists’ investment and reliance; Google needed Java’s API calls because of developers’ investment and reliance; software collections need to preserve and make available out-of-commerce titles because of the investment and reliance of the original users (who created proprietary files, imbued it with cultural value/meaning, or otherwise made the software a worthy subject of study) and the investment of scholars, teachers, and students (who find access to the software an important part of their work).

In other words, by the time a software program is being preserved and made accessible in a cultural heritage institution, its value to the collection is generally the result of other parties’ investment(s) in the program, not the investments of the copyright holder.

The public would suffer substantially if fair use did not protect software preservation.

Justice Breyer concludes his analysis of the market effect by observing that if fair use does not apply here, ownership of the Java API could serve as a “lock limiting the future creativity of new programs. Oracle alone would hold the key.” To the extent that such a lock is really the result of third-party investments in learning and using the Java platform, Breyer argues, it’s not consistent with the purposes of copyright to grant Oracle that kind of control. Instead, fair use should unlock this resource and enable the follow-on creativity that will result.

Here, again, the arguments for public benefit from software preservation are similar, and similarly strong. Consider again the case of an archival collection of digital documents that are only fully accessible when rendered by proprietary software. Without fair use, copyright would grant the maker of this software control over not just the software itself, but also the entire universe of digital resources that can only be fully accessed using that software. That level of control far exceeds the bounds of what copyright law typically protects, and it results in an unjustifiable limitation on public access and downstream creativity (in the form of teaching, research, and scholarship informed by access to the archive). Given the CONTU Report drafters’ admonition that copyright “should not grant anyone more economic power than is necessary to achieve the incentive to create,” fair use should apply, here, to constrain the market power of software vendors.

Takeaways for Cultural Heritage Institutions and Fair Use Generally

While the clearest implications of the Oracle opinion are for software, the principles deployed in this case are certainly susceptible to an even more generalized reading. At the level of cultural heritage uses generally, the most important takeaways are:

  • Fair use continues to play a crucial role in maintaining the central balance in the copyright system between the private incentive to create and distribute new works and the public interest in access and reuse of existing works. As cultural heritage institutions serve the public, they should understand and rely on their fair use rights.
  • Core public interest activities like teaching and research are especially worthy of fair use protection, and fair use doctrines like transformative use should be read with an eye to ensuring these activities are not unduly impeded by copyright law.
  • Specifically, when analyzing the purpose and character of a use for transformative use purposes, courts should avoid reductive or simplistic characterizations that result in an overly narrow scope of fair use or an overly broad scope of market power.

As an example of the possible influence of the Oracle opinion, imagine if it had been in place before the Georgia State course reserves case was decided. In that case, both the trial court and the 11th Circuit court of appeals gave a very hasty, reductive analysis of transformative use. In the space of a paragraph or less, the courts described the university’s use of excerpts from academic books as “mirror-image copies” that were assigned to be read, the same basic purpose as the original works themselves.

After Oracle, courts may not move so quickly through such an analysis. Instead, they may ask whether such a simple characterization would result in too narrow a field for fair use, and if so, whether a more nuanced characterization might result in a meaningful distinction between the user’s purpose and the original purpose of the work. This more probing analysis could lead courts to find transformative use more often in situations involving teaching, research, scholarship, and other activities where fair use is expected generally to have broader application. Given the effect of transformative use on the rest of the fair use analysis, the net result would be a meaningfully stronger fair use right for core fair users.


Oracle v. Google provides important fair use guidance for software preservation professionals, cultural heritage institutions, and all fair users. And luckily, it is very friendly, useful guidance that will serve software collections particularly well as they cope with the unique challenges posed by copyright in software. The pieces of the jigsaw puzzle of software copyright certainly fit together a little better today than they did before this opinion.

Preferred citation:

Butler, Brandon. (2021, June 3). Google v. Oracle Oral Argument: Quick Takeaways for Software Preservation. Software Preservation Network.